MFE: A Multimodal Hand Exoskeleton with Interactive Force, Pressure and Thermo-haptic Feedback

Abstract: Recent advancements in virtual reality and robotic teleoperation have greatly increased the variety of haptic information that must be conveyed to users. While existing haptic devices typically provide unimodal feedback to enhance situational awareness, a gap remains in their ability to deliver rich, multimodal sensory feedback encompassing force, pressure, and thermal sensations. To address this limitation, we present the Multimodal Feedback Exoskeleton (MFE), a hand exoskeleton designed to deliver hybrid haptic feedback. The MFE features 20 degrees of freedom for capturing hand pose. For force feedback, it employs an active mechanism capable of generating 3.5-8.1 N of pushing and pulling forces at the fingers' resting pose, enabling realistic interaction with deformable objects. The fingertips are equipped with flat actuators based on the electro-osmotic principle, providing pressure and vibration stimuli and achieving up to 2.47 kPa of contact pressure to render tactile sensations. For thermal feedback, the MFE integrates thermoelectric heat pumps capable of rendering temperatures from 10 to 55 degrees Celsius. We validated the MFE by integrating it into a robotic teleoperation system using the X-Arm 6 and Inspire Hand manipulator. In user studies, participants successfully recognized and manipulated deformable objects and differentiated remote objects with varying temperatures. These results demonstrate that the MFE enhances situational awareness, as well as the usability and transparency of robotic teleoperation systems.

• The paper demonstrates that combining bidirectional force, microfluidic pressure, and thermal feedback significantly enhances teleoperated object recognition and manipulation.
• It details a modular design with 20 DoF, microfluidic actuators delivering up to 2.47 kPa pressure, and a TEC module regulating temperature from 10° to 55°C via PID control.
• User studies revealed 100% shape recognition, high stiffness discrimination, and reduced manipulation failures, highlighting the benefits of multimodal feedback.

MFE: A Multimodal Feedback Exoskeleton for Enhanced Teleoperation

Introduction and Motivation

The paper "MFE: A Multimodal Hand Exoskeleton with Interactive Force, Pressure and Thermo-haptic Feedback" (2604.02820) proposes a low-cost, open-source hand exoskeleton capable of delivering hybrid haptic feedback—simultaneously providing active force, pressure, and temperature sensations. This responds to the persisting gap in commercial and open-source haptic devices, which often lack multi-modality, are prohibitively expensive, or do not reach the fidelity required for high-quality human teleoperation and data collection in embodied AI. The system is aimed at practical deployment scenarios including remote robotics operation, surgical telemanipulation, hazardous task execution, and tactile data acquisition for learning-based robotics.

Figure 1: The exoskeleton simultaneously captures finger positions and renders remote contact force, fingertip pressure, and palm temperature.

System Architecture and Multimodal Design

Force Feedback Subsystem

The exoskeleton features 20 DoFs for hand pose capture, using Dynamixel XL330-M288-T actuators coupled with 3D-printed linkages for direct force and position mapping. The system is unique in its bidirectional active force feedback—capable of both pushing and pulling actions per finger at up to 8.1 N output, accomplishing a degree of realism previously limited to high-end commercial exoskeletons while significantly reducing BOM costs.

Figure 2: Structural, canonical, and joint-tracked designs for the MFE exoskeleton, highlighting modular architecture and sensor integration.

Fingertip Pressure and Vibrotactile Feedback

Fingertip feedback is achieved with microfluidic actuators based on electro-osmotic pumps, enabling both static and dynamic (vibratile) pressure rendering without bulky mechanical pumps. These actuators provide localized stimuli with a maximum pressure of 2.47 kPa and 1.65 mm membrane deformation. Fabrication integrates high-density PCB electrodes, microfiltered fluidic paths, and elastomeric membranes, all controlled via high-voltage PWM drivers, ensuring tight response control and safety.

Figure 3: Construction, assembled view, individual components, and working principle of the microfluidic actuator module.

Thermal Feedback

Thermal cues are delivered via a palm-mounted thermoelectric heat pump (TEC), driven by closed-loop PID thermal control. The TEC achieves regulated output in the 10–55 °C range with active feedback and cooling fan integration to minimize actuation delay. On the robot side, a custom FPC-based flexible membrane array captures surface temperature for mapping to the human operator.

Integrated Teleoperation Pipeline

MFE is embedded in a full teleoperation ecosystem. The hand exoskeleton's joint encoder data—optionally increased with multi-joint Hall effect sensors—are mapped in real-time to a 6-DoF RH56BFX robot hand. Bilateral feedback includes contact force rendering via measured fingertip loads, pressure mapping with PWM-controlled microfluidic output, and temperature rendering using real-time sensor telemetry from the manipulated remote object.

Figure 4: Full technical pipeline of the MFE-based teleoperation system, spanning exoskeleton, robot hand, actuator and sensor arrays, and feedback loops.

Quantitative Benchmarks and Human Experiments

Device Characterization

Extensive parametric measurements quantify the capabilities of each feedback modality:

• Force feedback: Output ranges from 3.5 to 8.1 N, with low back-driving resistance (<0.5 N) and maximum measured linkage output of 4.5 N at rest pose—confirmed to be within human-operator safety envelopes.
• Microfluidic actuators: 2.47 kPa pressure, 1.65 mm deformation at peak, and rapid rise time (perceptible threshold reached within 0.1 s at 200 V), with pressure above human skin discrimination thresholds.
• Thermal actuation: TEC module covers 10–55 °C; steady-state transitions measurable within 3–6 s.

  Figure 5: Device-level characterization showing force, actuator output at varying voltage, and controllable temperature ranges.

Teleoperation User Study

A 10-subject study evaluated teleoperated object recognition, deformable object manipulation, and temperature discrimination using the MFE. Key results:

• Object recognition: 100% success in shape, 90% in stiffness with multimodal feedback; lower rates with unimodal feedback.
• Deformable manipulation: Significantly reduced content spillage (statistically significant) and decreased rate of failure with multimodal haptics, compared to force/pressure only or no feedback.
• Thermal cue perception: All subjects successfully ranked cup temperatures; typical recognition delay approximately 4 s with a minimal $\sim$5 °C difference threshold.

  Figure 6: Task setups for shape and stiffness discrimination, cup manipulation with deformable loads, and temperature ranking.

  

  Figure 7: Statistical evaluation of object manipulation performance under variant feedback conditions across users.

Implications, Limitations, and Future Directions

The MFE platform delivers a unique suite of open-source, modular, and low-cost multimodal haptic feedback. Strong human trial results demonstrate the clear superiority of hybrid feedback for teleoperation skill transfer and situational awareness. Notably, the device's cost/performance ratio outperforms established platforms, making it accessible for broad adoption in robotic skill learning, embodied AI data collection, VR, and medical telemanipulation contexts.

However, latency for thermal rendering (≈4 s) and dynamic response of pressure/force actuators (≈100–300 ms) remain limiting factors for high-speed manipulation tasks. Current designs also constrain temperature mapping to a spatially coarse, palm-located feedback. Future research is expected to focus on higher-resolution thermal arrays, enhanced actuator bandwidth, bus-based encoder networking, and direct use of MFE-mediated data for reinforcement or imitation learning in autonomous robot hands.

Conclusion

This work presents the first low-cost, open-source exoskeleton integrating bidirectional force, localized pressure, and thermal feedback for human-robot teleoperation. Quantitative device and user-level experiments illustrate substantially improved performance in object recognition, manipulation fidelity, and situational awareness. The MFE stands as a modular platform for further development in immersive telerobotics and as a research tool for multimodal data collection in AI-driven manipulation domains.

(2604.02820)